Ecosystems are complex owing to the fact that emergent properties like trophic structure and productivity depend on details related to lower-scale interactions among individuals. A key challenge is identifying how much individual-level detail is needed to predict patterns at the ecosystem level. We tested for the effect of individual herbivore body size on trophic interactions and consequent abundances of plant and herbivore trophic levels in a New England meadow ecosystem. Body size is an important determinant of vulnerability to predation and thus should influence the way individuals tradeoff time spent foraging against time spent avoiding contact with predators. Such tradeoffs can then influence the degree of damage herbivores inflict on their plant resources. We experimentally assigned field-caught grasshoppers to three distinct body size treatment groups (small, normal, and large) and crossed them with two spider predator treatments (spider present and absent) in a fully replicated design. We observed size-dependent differences in grasshopper survival and development. Moreover, predators caused grasshoppers to inflict greater damage to herbs and lesser damage to grasses relative to treatments without predators. However, there were no size-dependent differences in net damage level on grasses and herbs in either predator or no predator treatments owing to size-dependent compensation in grasshopper foraging effort. We thus conclude that in this ecosystem the foraging-predation risk tradeoff displayed by typical or average-sized herbivore is a sufficient amount of individual-level detail needed to explain ecosystem patterns.size-dependent predation risk ͉ herbivore-mediated trophic effects ͉ old-field ecosystem ͉ grasshoppers ͉ trait-mediated indirect effects E cosystems are paradigmatically complex. They contain many different components that interact directly and indirectly in integrated networks (1, 2). In such complex networks, higher scale system properties like trophic structure, nutrient fluxes, and productivity emerge from lower scale interactions and selection among components (1, 2). Furthermore, feedback loops in which higher scale properties modify lower-scale interactions cause new emergent properties to arise over time (1, 2). A central problem is identifying which lower scale processes should be included in theory aiming to predict higher-scale properties of ecosystems.Classical ecology (e.g., refs. 3-6) has approached this problem by assuming that it is sufficient to abstract lower scale details, such as interactions among individuals in populations, and characterize ecosystem function simply in terms of net changes in numbers or densities of individuals at the level of whole populations. Abstracting such individual-scale detail is reasonable if the effects of individual-level interactions attenuate on the time scale of changes in population density. However, the assumption that individual-scale detail can be safely abstracted is increasingly being called into question (7). Populations are effec...